script_core 0.2.1 → 0.2.6

data/ext/enterprise_script_service/mruby/mrbgems/mruby-compiler/core/node.h

@@ -51,6 +51,7 @@ enum node_type {
   NODE_IVAR,
   NODE_CONST,
   NODE_CVAR,
+  NODE_NVAR,
   NODE_NTH_REF,
   NODE_BACK_REF,
   NODE_MATCH,

data/ext/enterprise_script_service/mruby/mrbgems/mruby-compiler/core/parse.y

@@ -9,7 +9,6 @@
 #ifdef PARSER_DEBUG
 # define YYDEBUG 1
 #endif
-#define YYERROR_VERBOSE 1
 #define YYSTACK_USE_ALLOCA 1
 
 #include <ctype.h>
@@ -266,6 +265,7 @@ local_unnest(parser_state *p)
 static mrb_bool
 local_var_p(parser_state *p, mrb_sym sym)
 {
+  struct RProc *u;
   node *l = p->locals;
 
   while (l) {
@@ -276,6 +276,18 @@ local_var_p(parser_state *p, mrb_sym sym)
     }
     l = l->cdr;
   }
+
+  u = p->upper;
+  while (u && !MRB_PROC_CFUNC_P(u)) {
+    struct mrb_irep *ir = u->body.irep;
+    uint_fast16_t n = ir->nlocals;
+    const struct mrb_locals *v = ir->lv;
+    for (; n > 1; n --, v ++) {
+      if (v->name == sym) return TRUE;
+    }
+    if (MRB_PROC_SCOPE_P(u)) break;
+    u = u->upper;
+  }
   return FALSE;
 }
 
@@ -315,6 +327,24 @@ locals_node(parser_state *p)
   return p->locals ? p->locals->car : NULL;
 }
 
+static void
+nvars_nest(parser_state *p)
+{
+  p->nvars = cons(nint(0), p->nvars);
+}
+
+static void
+nvars_block(parser_state *p)
+{
+  p->nvars = cons(nint(-2), p->nvars);
+}
+
+static void
+nvars_unnest(parser_state *p)
+{
+  p->nvars = p->nvars->cdr;
+}
+
 /* (:scope (vars..) (prog...)) */
 static node*
 new_scope(parser_state *p, node *body)
@@ -649,6 +679,16 @@ new_cvar(parser_state *p, mrb_sym sym)
   return cons((node*)NODE_CVAR, nsym(sym));
 }
 
+/* (:nvar . a) */
+static node*
+new_nvar(parser_state *p, int num)
+{
+  int nvars = intn(p->nvars->car);
+
+  p->nvars->car = nint(nvars > num ? nvars : num);
+  return cons((node*)NODE_NVAR, nint(num));
+}
+
 /* (:const . a) */
 static node*
 new_const(parser_state *p, mrb_sym sym)
@@ -799,6 +839,13 @@ new_kw_arg(parser_state *p, mrb_sym kw, node *def_arg)
   return list3((node*)NODE_KW_ARG, nsym(kw), def_arg);
 }
 
+/* (:kw_rest_args . a) */
+static node*
+new_kw_rest_args(parser_state *p, node *a)
+{
+  return cons((node*)NODE_KW_REST_ARGS, a);
+}
+
 /* (:block_arg . a) */
 static node*
 new_block_arg(parser_state *p, node *a)
@@ -806,10 +853,41 @@ new_block_arg(parser_state *p, node *a)
   return cons((node*)NODE_BLOCK_ARG, a);
 }
 
+static node*
+setup_numparams(parser_state *p, node *a)
+{
+  int nvars = intn(p->nvars->car);
+  if (nvars > 0) {
+    int i;
+    mrb_sym sym;
+    // m || opt || rest || tail
+    if (a && (a->car || (a->cdr && a->cdr->car) || (a->cdr->cdr && a->cdr->cdr->car) || (a->cdr->cdr->cdr->cdr && a->cdr->cdr->cdr->cdr->car))) {
+      yyerror(p, "ordinary parameter is defined");
+    }
+    else if (p->locals) {
+      /* p->locals should not be NULL unless error happens before the point */
+      node* args = 0;
+      for (i = nvars; i > 0; i--) {
+        char buf[3];
+
+        buf[0] = '_';
+        buf[1] = i+'0';
+        buf[2] = '\0';
+        sym = intern_cstr(buf);
+        args = cons(new_arg(p, sym), args);
+        p->locals->car = cons(nsym(sym), p->locals->car);
+      }
+      a = new_args(p, args, 0, 0, 0, 0);
+    }
+  }
+  return a;
+}
+
 /* (:block arg body) */
 static node*
 new_block(parser_state *p, node *a, node *b)
 {
+  a = setup_numparams(p, a);
   return list4((node*)NODE_BLOCK, locals_node(p), a, b);
 }
 
@@ -1249,7 +1327,6 @@ heredoc_end(parser_state *p)
   p->parsing_heredoc = p->parsing_heredoc->cdr;
   if (p->parsing_heredoc == NULL) {
     p->lstate = EXPR_BEG;
-    p->cmd_start = TRUE;
     end_strterm(p);
     p->lex_strterm = p->lex_strterm_before_heredoc;
     p->lex_strterm_before_heredoc = NULL;
@@ -1265,7 +1342,8 @@ heredoc_end(parser_state *p)
 
 %}
 
-%pure-parser
+%define parse.error verbose
+%define api.pure
 %parse-param {parser_state *p}
 %lex-param {parser_state *p}
 
@@ -1334,6 +1412,7 @@ heredoc_end(parser_state *p)
 %token <nd>  tSTRING tSTRING_PART tSTRING_MID
 %token <nd>  tNTH_REF tBACK_REF
 %token <num> tREGEXP_END
+%token <num> tNUMPARAM
 
 %type <nd> singleton string string_fragment string_rep string_interp xstring regexp
 %type <nd> literal numeric cpath symbol
@@ -1466,11 +1545,13 @@ top_stmt        : stmt
                 | keyword_BEGIN
                     {
                       $<nd>$ = local_switch(p);
+                      nvars_block(p);
                     }
                   '{' top_compstmt '}'
                     {
                       yyerror(p, "BEGIN not supported");
                       local_resume(p, $<nd>2);
+                      nvars_unnest(p);
                       $$ = 0;
                     }
                 ;
@@ -1668,6 +1749,7 @@ block_command   : block_call
 cmd_brace_block : tLBRACE_ARG
                     {
                       local_nest(p);
+                      nvars_nest(p);
                     }
                   opt_block_param
                   compstmt
@@ -1675,6 +1757,7 @@ cmd_brace_block : tLBRACE_ARG
                     {
                       $$ = new_block(p, $3, $4);
                       local_unnest(p);
+                      nvars_unnest(p);
                     }
                 ;
 
@@ -1889,6 +1972,10 @@ lhs             : variable
                       backref_error(p, $1);
                       $$ = 0;
                     }
+                | tNUMPARAM
+                    {
+                      yyerror(p, "can't assign to numbered parameter");
+                    }
                 ;
 
 cname           : tIDENTIFIER
@@ -2198,6 +2285,30 @@ paren_args      : '(' opt_call_args ')'
                     {
                       $$ = $2;
                     }
+                | '(' tDOT3 rparen
+                    {
+#if 1
+                      mrb_sym r = mrb_intern_lit(p->mrb, "*");
+                      mrb_sym b = mrb_intern_lit(p->mrb, "&");
+                      if (local_var_p(p, r)  && local_var_p(p, b)) {
+                        $$ = cons(list1(new_splat(p, new_lvar(p, r))),
+                                  new_block_arg(p, new_lvar(p, b)));
+                      }
+#else
+                      mrb_sym r = mrb_intern_lit(p->mrb, "*");
+                      mrb_sym k = mrb_intern_lit(p->mrb, "**");
+                      mrb_sym b = mrb_intern_lit(p->mrb, "&");
+                      if (local_var_p(p, r) && local_var_p(p, k) && local_var_p(p, b)) {
+                        $$ = cons(list2(new_splat(p, new_lvar(p, r)),
+                                        new_kw_hash(p, list1(cons(new_kw_rest_args(p, 0), new_lvar(p, k))))),
+                                  new_block_arg(p, new_lvar(p, b)));
+                      }
+#endif
+                      else {
+                        yyerror(p, "unexpected argument forwarding ...");
+                        $$ = 0;
+                      }
+                    }
                 ;
 
 opt_paren_args  : none
@@ -2279,7 +2390,7 @@ opt_block_arg   : comma block_arg
                 ;
 
 comma           : ','
-                | ','  heredoc_bodies
+                | ','  opt_nl heredoc_bodies
                 ;
 
 args            : arg
@@ -2330,6 +2441,10 @@ primary         : literal
                 | heredoc
                 | var_ref
                 | backref
+                | tNUMPARAM
+                    {
+                      $$ = new_nvar(p, $1);
+                    }
                 | tFID
                     {
                       $$ = new_fcall(p, $1, 0);
@@ -2482,6 +2597,7 @@ primary         : literal
                       if (p->in_def || p->in_single)
                         yyerror(p, "class definition in method body");
                       $<nd>$ = local_switch(p);
+                      nvars_block(p);
                     }
                   bodystmt
                   keyword_end
@@ -2489,6 +2605,7 @@ primary         : literal
                       $$ = new_class(p, $2, $3, $5);
                       SET_LINENO($$, $1);
                       local_resume(p, $<nd>4);
+                      nvars_unnest(p);
                     }
                 | keyword_class
                   tLSHFT expr
@@ -2499,6 +2616,7 @@ primary         : literal
                   term
                     {
                       $<nd>$ = cons(local_switch(p), nint(p->in_single));
+                      nvars_block(p);
                       p->in_single = 0;
                     }
                   bodystmt
@@ -2507,6 +2625,7 @@ primary         : literal
                       $$ = new_sclass(p, $3, $7);
                       SET_LINENO($$, $1);
                       local_resume(p, $<nd>6->car);
+                      nvars_unnest(p);
                       p->in_def = $<num>4;
                       p->in_single = intn($<nd>6->cdr);
                     }
@@ -2516,6 +2635,7 @@ primary         : literal
                       if (p->in_def || p->in_single)
                         yyerror(p, "module definition in method body");
                       $<nd>$ = local_switch(p);
+                      nvars_block(p);
                     }
                   bodystmt
                   keyword_end
@@ -2523,6 +2643,7 @@ primary         : literal
                       $$ = new_module(p, $2, $4);
                       SET_LINENO($$, $1);
                       local_resume(p, $<nd>3);
+                      nvars_unnest(p);
                     }
                 | keyword_def fname
                     {
@@ -2532,6 +2653,7 @@ primary         : literal
                     {
                       p->in_def++;
                       $<nd>$ = local_switch(p);
+                      nvars_block(p);
                     }
                   f_arglist
                   bodystmt
@@ -2540,6 +2662,7 @@ primary         : literal
                       $$ = new_def(p, $2, $5, $6);
                       SET_LINENO($$, $1);
                       local_resume(p, $<nd>4);
+                      nvars_unnest(p);
                       p->in_def--;
                       p->cmdarg_stack = $<stack>3;
                     }
@@ -2554,6 +2677,7 @@ primary         : literal
                       p->in_single++;
                       p->lstate = EXPR_ENDFN; /* force for args */
                       $<nd>$ = local_switch(p);
+                      nvars_block(p);
                     }
                   f_arglist
                   bodystmt
@@ -2562,6 +2686,7 @@ primary         : literal
                       $$ = new_sdef(p, $2, $5, $7, $8);
                       SET_LINENO($$, $1);
                       local_resume(p, $<nd>6);
+                      nvars_unnest(p);
                       p->in_single--;
                       p->cmdarg_stack = $<stack>4;
                     }
@@ -2829,6 +2954,7 @@ lambda_body     : tLAMBEG compstmt '}'
 do_block        : keyword_do_block
                     {
                       local_nest(p);
+                      nvars_nest(p);
                     }
                   opt_block_param
                   bodystmt
@@ -2836,6 +2962,7 @@ do_block        : keyword_do_block
                     {
                       $$ = new_block(p,$3,$4);
                       local_unnest(p);
+                      nvars_unnest(p);
                     }
                 ;
 
@@ -2906,6 +3033,7 @@ method_call     : operation paren_args
 brace_block     : '{'
                     {
                       local_nest(p);
+                      nvars_nest(p);
                       $<num>$ = p->lineno;
                     }
                   opt_block_param
@@ -2914,10 +3042,12 @@ brace_block     : '{'
                       $$ = new_block(p,$3,$4);
                       SET_LINENO($$, $<num>2);
                       local_unnest(p);
+                      nvars_unnest(p);
                     }
                 | keyword_do
                     {
                       local_nest(p);
+                      nvars_nest(p);
                       $<num>$ = p->lineno;
                     }
                   opt_block_param
@@ -2926,6 +3056,7 @@ brace_block     : '{'
                       $$ = new_block(p,$3,$4);
                       SET_LINENO($$, $<num>2);
                       local_unnest(p);
+                      nvars_unnest(p);
                     }
                 ;
 
@@ -3192,6 +3323,10 @@ var_lhs         : variable
                     {
                       assignable(p, $1);
                     }
+                | tNUMPARAM
+                    {
+                      yyerror(p, "can't assign to numbered parameter");
+                    }
                 ;
 
 var_ref         : variable
@@ -3270,6 +3405,24 @@ f_arglist       : '(' f_args rparen
                       p->lstate = EXPR_BEG;
                       p->cmd_start = TRUE;
                     }
+                | '(' tDOT3 rparen
+                    {
+#if 1
+                      /* til real keyword args implemented */
+                      mrb_sym r = mrb_intern_lit(p->mrb, "*");
+                      mrb_sym b = mrb_intern_lit(p->mrb, "&");
+                      local_add_f(p, r);
+                      $$ = new_args(p, 0, 0, r, 0,
+                                    new_args_tail(p, 0, 0, b));
+#else
+                      mrb_sym r = mrb_intern_lit(p->mrb, "*");
+                      mrb_sym k = mrb_intern_lit(p->mrb, "**");
+                      mrb_sym b = mrb_intern_lit(p->mrb, "&");
+                      local_add_f(p, r); local_add_f(p, k);
+                      $$ = new_args(p, 0, 0, r, 0,
+                                    new_args_tail(p, 0, new_kw_rest_args(p, nsym(k)), b));
+#endif
+                    }
                 | f_args term
                     {
                       $$ = $1;
@@ -3333,11 +3486,11 @@ kwrest_mark     : tPOW
 
 f_kwrest        : kwrest_mark tIDENTIFIER
                     {
-                      $$ = cons((node*)NODE_KW_REST_ARGS, nsym($2));
+                      $$ = new_kw_rest_args(p, nsym($2));
                     }
                 | kwrest_mark
                     {
-                      $$ = cons((node*)NODE_KW_REST_ARGS, 0);
+                      $$ = new_kw_rest_args(p, 0);
                     }
                 ;
 
@@ -3452,6 +3605,11 @@ f_bad_arg       : tCONSTANT
                       yyerror(p, "formal argument cannot be a class variable");
                       $$ = 0;
                     }
+                | tNUMPARAM
+                    {
+                      yyerror(p, "formal argument cannot be a numbered parameter");
+                      $$ = 0;
+                    }
                 ;
 
 f_norm_arg      : f_bad_arg
@@ -3613,24 +3771,28 @@ assocs          : assoc
                       $$ = list1($1);
                       NODE_LINENO($$, $1);
                     }
-                | assocs ',' assoc
+                | assocs comma assoc
                     {
                       $$ = push($1, $3);
                     }
                 ;
 
+label_tag       : tLABEL_TAG
+                | tLABEL_TAG heredoc_bodies
+                ;
+
 assoc           : arg tASSOC arg
                     {
                       void_expr_error(p, $1);
                       void_expr_error(p, $3);
                       $$ = cons($1, $3);
                     }
-                | tIDENTIFIER tLABEL_TAG arg
+                | tIDENTIFIER label_tag arg
                     {
                       void_expr_error(p, $3);
                       $$ = cons(new_sym(p, $1), $3);
                     }
-                | string_fragment tLABEL_TAG arg
+                | string_fragment label_tag arg
                     {
                       void_expr_error(p, $3);
                       if ($1->car == (node*)NODE_DSTR) {
@@ -3643,7 +3805,7 @@ assoc           : arg tASSOC arg
                 | tDSTAR arg
                     {
                       void_expr_error(p, $2);
-                      $$ = cons(cons((node*)NODE_KW_REST_ARGS, 0), $2);
+                      $$ = cons(new_kw_rest_args(p, 0), $2);
                     }
                 ;
 
@@ -3707,7 +3869,7 @@ term            : ';' {yyerrok;}
 
 nl              : '\n'
                     {
-                      p->lineno++;
+                      p->lineno += $<num>1;
                       p->column = 0;
                     }
                 ;
@@ -3866,6 +4028,27 @@ static void pushback(parser_state *p, int c);
 static mrb_bool peeks(parser_state *p, const char *s);
 static mrb_bool skips(parser_state *p, const char *s);
 
+static inline int
+nextc0(parser_state *p)
+{
+  int c;
+#ifndef MRB_DISABLE_STDIO
+  if (p->f) {
+    if (feof(p->f)) return -1;
+    c = fgetc(p->f);
+    if (c == EOF) return -1;
+  }
+  else
+#endif
+    if (!p->s || p->s >= p->send) {
+      return -1;
+    }
+    else {
+      c = (unsigned char)*p->s++;
+    }
+  return c;
+}
+
 static inline int
 nextc(parser_state *p)
 {
@@ -3880,24 +4063,19 @@ nextc(parser_state *p)
     cons_free(tmp);
   }
   else {
-#ifndef MRB_DISABLE_STDIO
-    if (p->f) {
-      if (feof(p->f)) goto eof;
-      c = fgetc(p->f);
-      if (c == EOF) goto eof;
-    }
-    else
-#endif
-      if (!p->s || p->s >= p->send) {
-        goto eof;
-      }
-      else {
-        c = (unsigned char)*p->s++;
-      }
+    c = nextc0(p);
+    if (c < 0) goto eof;
   }
   if (c >= 0) {
     p->column++;
   }
+  if (c == '\r') {
+    const int lf = nextc0(p);
+    if (lf == '\n') {
+      return '\n';
+    }
+    if (lf > 0) pushback(p, lf);
+  }
   return c;
 
   eof:
@@ -4679,6 +4857,7 @@ static int
 parser_yylex(parser_state *p)
 {
   int32_t c;
+  int nlines = 1;
   int space_seen = 0;
   int cmd_state;
   enum mrb_lex_state_enum last_state;
@@ -4736,6 +4915,7 @@ parser_yylex(parser_state *p)
       break;
     }
     if (p->parsing_heredoc != NULL) {
+      pylval.num = nlines;
       return '\n';
     }
     while ((c = nextc(p))) {
@@ -4745,13 +4925,13 @@ parser_yylex(parser_state *p)
         space_seen = 1;
         break;
       case '#': /* comment as a whitespace */
-        pushback(p, '#');
-        p->lineno++;
-        goto retry;
+        skip(p, '\n');
+        nlines++;
+        break;
       case '.':
         if (!peek(p, '.')) {
           pushback(p, '.');
-          p->lineno++;
+          p->lineno+=nlines; nlines=1;
           goto retry;
         }
         pushback(p, c);
@@ -4759,7 +4939,7 @@ parser_yylex(parser_state *p)
       case '&':
         if (peek(p, '.')) {
           pushback(p, '&');
-          p->lineno++;
+          p->lineno+=nlines; nlines=1;
           goto retry;
         }
         pushback(p, c);
@@ -4775,6 +4955,7 @@ parser_yylex(parser_state *p)
   normal_newline:
     p->cmd_start = TRUE;
     p->lstate = EXPR_BEG;
+    pylval.num = nlines;
     return '\n';
 
   case '*':
@@ -4857,7 +5038,7 @@ parser_yylex(parser_state *p)
             c = nextc(p);
           } while (!(c < 0 || ISSPACE(c)));
           if (c != '\n') skip(p, '\n');
-          p->lineno++;
+          p->lineno+=nlines; nlines=1;
           p->column = 0;
           goto retry;
         }
@@ -5579,7 +5760,7 @@ parser_yylex(parser_state *p)
   case '\\':
     c = nextc(p);
     if (c == '\n') {
-      p->lineno++;
+      p->lineno+=nlines; nlines=1;
       p->column = 0;
       space_seen = 1;
       goto retry; /* skip \\n */
@@ -5861,6 +6042,39 @@ parser_yylex(parser_state *p)
         result = tIVAR;
       break;
 
+    case '_':
+      if (toklen(p) == 2 && ISDIGIT(tok(p)[1]) && p->nvars) {
+        int n = tok(p)[1] - '0';
+        int nvar;
+
+        if (n > 0) {
+          node *nvars = p->nvars->cdr;
+
+          while (nvars) {
+            nvar = intn(nvars->car);
+            if (nvar == -2) break; /* top of the scope */
+            if (nvar > 0) {
+              yywarning(p, "numbered parameter used in outer block");
+              break;
+            }
+            nvars->car = nint(-1);
+            nvars = nvars->cdr;
+          }
+          nvar = intn(p->nvars->car);
+          if (nvar == -1) {
+            yywarning(p, "numbered parameter used in inner block");
+          }
+          if (nvar >= -1) {
+            pylval.num = n;
+            p->lstate = EXPR_END;
+            return tNUMPARAM;
+          }
+          else {
+            yywarning(p, "identifier for numbered parameter; consider another name");
+          }
+        }
+      }
+      /* fall through */
     default:
       if (toklast(p) == '!' || toklast(p) == '?') {
         result = tFID;
@@ -5876,6 +6090,15 @@ parser_yylex(parser_state *p)
           else {
             pushback(p, c);
           }
+          if ((c = nextc(p)) == '=' && !peek(p, '~') && !peek(p, '>') &&
+              (!peek(p, '=') || (peek_n(p, '>', 1)))) {
+            result = tIDENTIFIER;
+            tokadd(p, c);
+            tokfix(p);
+          }
+          else {
+            pushback(p, c);
+          }
         }
         if (result == 0 && ISUPPER(tok(p)[0])) {
           result = tCONSTANT;
@@ -5982,7 +6205,7 @@ parser_init_cxt(parser_state *p, mrbc_context *cxt)
   }
   p->capture_errors = cxt->capture_errors;
   p->no_optimize = cxt->no_optimize;
-  p->on_eval = cxt->on_eval;
+  p->upper = cxt->upper;
   if (cxt->partial_hook) {
     p->cxt = cxt;
   }
@@ -6401,7 +6624,7 @@ dump_args(mrb_state *mrb, node *n, int offset)
 static const char*
 str_dump(mrb_state *mrb, const char *str, int len)
 {
-  mrb_int ai = mrb_gc_arena_save(mrb);
+  int ai = mrb_gc_arena_save(mrb);
   mrb_value s;
 # if INT_MAX > MRB_INT_MAX / 4
   /* check maximum length with "\xNN" charactor */
@@ -6843,6 +7066,10 @@ mrb_parser_dump(mrb_state *mrb, node *tree, int offset)
     printf("NODE_CVAR %s\n", mrb_sym_name(mrb, sym(tree)));
     break;
 
+  case NODE_NVAR:
+    printf("NODE_NVAR %d\n", intn(tree));
+    break;
+
   case NODE_CONST:
     printf("NODE_CONST %s\n", mrb_sym_name(mrb, sym(tree)));
     break;